Vol. 60, No. 1INFECTION AND IMMUNITY, Jan. 1992, p. 117-1230019-9567/92/010117-07$02.00/0Copyright © 1992, American Society for Microbiology

Siderophore Production and Membrane Alterations byBordetella pertussis in Response to Iron Starvation

LISA-ANNE AGIATO* AND DAVID W. DYERDepartment of Microbiology, State University ofNew York at Buffalo,

Buffalo, New York 14214

Received 7 June 1991/Accepted 3 October 1991

Bordetella pertussis was grown in iron (Fe)-free defined medium to limit the growth of the organism.Doubling times of the Fe-starved organism increased by approximately 1 h, and a 40% reduction in the finalextent of growth in Fe-depleted medium was observed. Under these conditions, a hydroxamate siderophorenamed bordetellin was secreted by B. pertussis. Lactoferrin and transferrin supported growth of B. pertussiseven when the protein was sequestered inside dialysis tubing. This suggested that binding of lactoferrin andtransferrin to B. pertussis was not essential and that bordetellin production plays a major role in Fe uptake.Solid-phase dot blot assays indicated weak binding of lactoferrin to the cell surface, consistent with previousreports of a lactoferrin receptor. Three new proteins of 97, 77, and 63 kDa were synthesized in response to Festarvation. Fe-inducible proteins of 103, 72, 24, 21, and 18 kDa were also observed. The synthesis oflipopolysaccharide was also altered by Fe availability.

Bordetella pertussis is the causative agent of whoopingcough. This gram-negative coccobacillus is transmittedthrough the inhalation of infected droplets. The organismestablishes infection by colonizing the upper respiratorymucosa. This initial stage of infection is termed the catarrhalphase and lasts approximately 2 weeks. As the infectionprogresses, the organism moves down the respiratory tree,causing inflammation. It is during this paroxysmal stage ofinfection that the characteristic cough associated with thedisease is present.

Iron (Fe) is a growth requirement for virtually all microbes(36). In animals, most Fe is found intracellularly, seques-tered in ferritin or bound in heme compounds. In the humanbody, extracellular Fe" is sequestered from microbes bybinding to proteins such as lactoferrin (LF) and transferrin(TF). These mechanisms keep the level of free Fe below theminimum levels (0.4 to 1 ,uM) required for the growth ofmost microbes (36). These nonspecific mechanisms for re-stricting bacterial growth have been termed nutritional im-munity (36). Pathogenic bacteria must be capable of remov-ing the Fe from host Fe-binding compounds if they are toinitiate and sustain infection. Several mechanisms exist bywhich bacteria remove Fe from host compounds. Manymicroorganisms such as Escherichia coli, Pseudomonasspp., and Shigella spp. secrete low-molecular-weight Fe-chelating compounds termed siderophores (20). Sidero-phores typically bind ferric iron; the siderophores producedby pathogenic bacteria are commonly able to remove Fe3+from LF and TF (19). The ferri-siderophore complex is thenbound by a bacterial surface receptor and processed toremove the Fe3+ ion for use by the organism (19).Other microbes, such as the pathogenic Neisseria spp.,

possess receptors for host Fe-binding proteins (27, 35).These pathogens bind TF and LF to separate outer mem-brane receptors, and the Fe is then removed from theprotein. Haemophilus influenzae has also been reported toacquire Fe from TF by a siderophore-independent mecha-

* Corresponding author.

nism and the acquisition of Fe from the protein requiredcell-to-protein contact (18).Few reports have appeared in the literature concerning the

Fe uptake system of B. pertussis. Redhead et al. (24)suggested that B. pertussis synthesized a receptor for hostFe-binding proteins in response to Fe starvation. Theseresearchers concluded that a single cell surface receptorbound both human TF and LF, unlike the pathogenic Neis-seria spp. which have separate receptors. In these studies,siderophore production was not observed (24). Gorringe etal. (9) reported that B. pertussis secreted a hydroxamatesiderophore in response to low Fe availability. This reportconcluded that B. pertussis utilized a dual mechanism of Feuptake, making use of both the hydroxamate siderophoreand the receptor described by Redhead et al. (24). Morerecently, Redhead and Hill reported the removal of Fe3"from TF by a receptor-mediated mechanism, but the puta-tive receptor was not identified (23).

This study provides new information on the Fe sourcescapable of supporting growth of the organism. We havecollected evidence that a siderophore-mediated Fe uptakesystem plays a major role in Fe3" acquisition from LF andTF. We examined the membrane protein and lipopolysac-charide (LPS) profiles of cells grown under high- and low-Feconditions and report the changes observed in synthesis dueto Fe availability.

MATERIALS AND METHODS

Bacterial strains and growth media. For most experiments,B. pertussis DBP2, a spontaneous streptomycin-resistantderivative of strain Tohama I, was used. In certain in-stances, B. pertussis 18323 and BB114 were also examined.Bacterial strains were maintained on Stainer-Scholte (30)agar plates containing 10% defibrinated sheep blood (CraneLaboratory, Syracuse, N.Y.). Hemolytic (vir+) bacteriawere used in all experiments. Plates were incubated in a 37°Cand 5% CO2 environment in a Napco model 6100 CO2incubator (Napco Scientific Co., Tualatin, Oreg.). Brothcultures were grown in Stainer-Scholte medium (SSM) or inChelex-treated defined medium (CDM) (37). Broth cultures

117

on August 29, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

118 AGIATO AND DYER

were shaken at 250 rpm at 37°C in a Series 25 incubatorshaker (New Brunswick Scientific, Edison, N.J.). In certainexperiments, streptomycin was included in the medium at aconcentration of 200 jig/ml in broth and 400 ,ug/ml in bloodplates. All reagents were purchased from Sigma ChemicalCo., St. Louis, Mo., unless otherwise noted.

Iron depletion of growth media. All glassware was soakedovernight in 3 N nitric acid and rinsed extensively withdistilled, deionized water. Disposable plasticware was usedwhenever possible to eliminate Fe contamination. All mediawere depleted of Fe by treatment with the cation exchangeresin Chelex 100 (Bio-Rad Laboratories, Richmond, Calif.)as previously described by West and Sparling (37). All waterused for preparing Fe-free media had a resistivity of at least15 MQ1. The amount of residual Fe in Chelex-treated SSM(CSSM) was determined with a ferrozine reagent (31). Theaverage Fe concentration from several batches of CSSM wascalculated to be 0.3 ,uM. The Fe concentration was alsomeasured by atomic absorption with a Perkin-Elmer 1100Batomic absorption spectrophotometer equipped with anHGA 700 graphite furnace and used according to the manu-facturer's recommendations (Perkin-Elmer, Oak Brook,Ill.). Fe(NO3)3 (0.2 ,ug/ml) in 1% HNO3 was used as thestandard. By this method, the Fe content of CSSM wasdetermined to be 0.5 ,uM. This amount of Fe is the minimumlevel required for growth of most gram-negative organisms(36).Growth of B. pertussis with nonprotein iron. Strain DBP2

was grown overnight in CSSM without added Fe. The cellswere harvested by centrifugation at 4,000 x g for 15 min atroom temperature and washed twice in fresh CSSM, and thecells were suspended in 20 ml of CSSM. A 5% inoculum wasadded to fresh CSSM, and various Fe sources were added toa final concentration of 10 ,M Fe. Growth was measured byoptical density in a Klett-Summerson colorimeter equippedwith a green filter (Klett Mfg. Co., Long Island City, N.Y.).Heme was dissolved in 10 mM NaOH, filter sterilized, and

used at a final concentration of 10 ,iM. Ferric dicitrate andferric pyrophosphate were made by combining sodium cit-rate or sodium pyrophosphate and ferric nitrate in a 10:1molar ratio (citrate or pyrophosphate to Fe). All Fe sourceswere prepared immediately before use, filter sterilized, andadded to the medium at a final concentration of 10 ,uM Fe.Growth with TF and LF. Growth curves in CDM were

performed to determine whether human LF and TF wouldsupport the growth of B. pertussis. LF and TF were dis-solved at 16 mg/ml in a solution of 50 mM Tris (pH 7.5), 150mM NaCl, and 20 mM NaHCO3 to give a final proteinconcentration of 200 ,uM. The proteins were 30% saturatedwith Fe by the addition of 120 ,uM ferric dicitrate and thenincubated at 37°C for 1 h. Citrate and unbound Fe wereremoved by overnight dialysis against 4 liters of a solution of50 mM Tris (pH 7.5), 150 mM NaCl, and 20 mM NaHCO3with two changes of buffer. The proteins were then filtersterilized. The protein concentration was determined byusing Bio-Rad protein dye concentrate with bovine serumalbumin as the reference. The amount of Fe bound to theproteins was measured spectrophotometrically using theextinction coefficients of the Fe-saturated forms of theproteins: human LF, e" = 0.51; human TF, 64% = 0.6 (1).NaHCO3 (10 mM) was added to the medium containing TFand LF to maintain strong binding of Fe3+ to the protein(10). The proteins were added to the medium to a finalconcentration of 10 ,uM Fe.

In some experiments, TF and LF were labeled with 1251using lodo-beads (Pierce, Rockford, Ill.) according to the

manufacturer's specifications. Nal (50 mM) was added to 5ml of TF or LF (at a concentration of 16 mg/ml) andincubated for 30 min at 37°C. This step was added tosuppress nonspecific binding of 125I to these proteins. Theprotein was then dialyzed against 4 liters of a solution of 50mM Tris (pH 7.5), 150 mM NaCl, and 20 mM NaHCO3overnight to remove any unbound Nal. Two lodo-beadswere washed in 5 ml of a solution of 50 mM Tris (pH 7.5), 150mM NaCl, and 20 mM NaHCO3. The beads were thenincubated in 500 ,ul of buffer containing 100 p.Ci of 1251 for 15min at room temperature. TF or LF was then added toIodo-beads and allowed to incubate at room temperature for15 min. The reaction was stopped by removing the proteinfrom the beads. The protein was dialyzed as described abovewith additional dialysis against 2 liters of CDM overnightwith two changes of buffer. To ensure that the protein hadnot been damaged during the iodination procedure, thelabeled protein was run on 7.5% polyacrylamide gels (13)and silver stained (38). The gel was dried and used to exposeKodak XAR film (Kodak, Rochester, N.Y.) for 16 h.

In order to determine whether cell-to-protein contact wasnecessary for growth with TF and LF, the radiolabeledprotein was sequestered inside Spectra/Por 1 dialysis tubingwith a molecular mass cutoff of 6,000 to 8,000 Da (SpectrumMedical Industries, Los Angeles, Calif.). The tubing waslong enough so that both ends were closed with a singleSpectra/Por closure (Spectrum Medical Industries) at themouth of the flask. Neither end of the dialysis tubing wasimmersed in the culture medium. This ensured that theprotein contained in the dialysis bag would not leak into themedium from an incompletely closed dialysis clip. Proteinwas pushed to the middle of the tubing which ran along thebottom of the flask and was completely immersed in CDM.Control flasks containing sterile CDM in the tubing wereincluded to ensure that Fe was not introduced on the dialysistubing and that the tubing itself did not interfere with growth.

Siderophore assays. B. pertussis was grown to late logphase in CSSM and in CSSM plus 10 p.M added Fe. Cellswere removed by centrifugation, and the supernatants werefilter sterilized through a 0.45-p.m-pore-size nylon membrane(Nalgene Co., Rochester, N.Y.). The chrome azurol S(CAS) assay of Schwyn and Neilands (28) was performedessentially as described by these researchers using Desferalmesylate (CIBA Pharmaceuticals, Summit, N.J.) as a refer-ence Fe-binding compound. The Csaky assay (7), whichdetects the presence of hydroxamic acids, was carried out byusing hydroxylamine to generate a standard curve. In thisassay, we used a 30-min acid hydrolysis step instead of the4-h step originally suggested (7). Longer hydrolysis timessignificantly decreased detection of the hydroxylamine stan-dard and any hydroxamate present in the samples. Culturesupernatants were also tested for the presence of phenolatesusing the Arnow assay (2).The Fe-binding ability of an excreted product in the

culture supernatant was also assayed spectrophotometri-cally. Ferric nitrate (final concentration, 100 ,uM) was addedto filter-sterilized culture supernatant from cells grown inCSSM and in CSSM plus 10 p.M FeSO4. A differenceabsorption scan was performed on the supernatant from theFe-starved cultures using the Fe-replete culture supernatantas the reference. A Perkin-Elmer Lambda 3 UV-Vis spec-trophotometer equipped with a chart recorder was used toscan the supernatant from 600 to 390 nm.

Receptor assay for human LF and TF. Dot blot assays wereperformed as described previously (27) using LF and TFconjugated to alkaline phosphatase (AP) (Jackson Immuno-

INFECT. IMMUN.

on August 29, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

SIDEROPHORE PRODUCTION OF B. PERTUSSIS 119

chemicals, Avondale, Pa.). Briefly, cells were grown to latelog phase in CDM and CDM plus 10 ,uM Fe. A 1.0-ml aliquotwas removed from the culture and adjusted to an opticaldensity of 100 Klett units. A 100-1±l aliquot was filtered ontoBA85 nitrocellulose (Schleicher and Schuell, Keene, N.H.),using a Mini-fold II apparatus (Schleicher and Schuell). Thedots were allowed to air dry. Nonspecific protein bindingsites on the nitrocellulose were blocked with Tris-bufferedsaline (50 mM Tris [pH 7.5], 150 mM NaCl) containing 0.5%skim milk. Cells were probed with 2 ,ug of TF-AP per ml infresh blocking buffer; LF-AP (2 ,ug/ml) was diluted in block-ing buffer in which the NaCl concentration had been in-creased to 500 mM to suppress nonspecific binding (4).Phosphatase activity was detected by using a Bio-Rad Alka-line Phosphatase Conjugate Substrate Kit. Neisseria menin-gitidis, which has been shown to possess TF and LFreceptors (27, 35), was used as a positive control for receptoractivity. Control dots not incubated with the conjugate wereincluded to ensure that no false-positive reactions occurredas a result of endogenous AP activity.Membrane alterations in response to iron availability.

Strain DBP2 was grown to late log phase in CSSM andCSSM plus Fe. Cells were collected by centrifugation for 10min at 10,000 x g, and membranes were prepared essentiallyas described by Redhead (22). After centrifugation, the cellpellet was suspended in 5 ml of 10 mM HEPES (N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid) (pH 8.0),and the cells were disrupted by passage through a Frenchpressure cell (SLM Instruments, Urbana, Ill.) at 16,000lb/in2. Unbroken cells were removed by centrifugation at10,000 x g for 10 min. Crude membrane fractions werecollected by ultracentrifugation in a Beckman 70.1 Ti rotor at100,000 x g for 1 h. These crude membranes were washed in10 mM HEPES (pH 8.0) and pelleted by ultracentrifugationas described above. Enrichment for outer membranes wasachieved by extraction with 2% Triton X-100 in 10 mMHEPES (pH 8.0), followed by ultracentrifugation for 1 h at100,000 x g. The Triton X-100-insoluble pellet was collectedand resuspended in 10 mM HEPES (pH 8.0) (22). Membraneprotein concentrations were determined by a modifiedLowry method (15). Membrane proteins were separated byelectrophoresis through 7.5 and 15% polyacrylamide-sodiumdodecyl sulfate (SDS) gels (13) using gels (13 by 16 cm) in aHoeffer Scientific gel box. Proteins were visualized by silverstaining (38).

B. pertussis LPS was also examined. Membrane sampleswere digested with proteinase K (20 ,ug/mg of membraneprotein) for 1 h at 37°C (12). The digested samples wereelectrophoresed through 14% acrylamide gels containing18% urea (14). LPS was stained by silver staining withperiodic acid oxidation (34). LPS from Salmonella minne-sota LPS mutants Ra, Rc, and R, were used as molecularsize markers. The molecular masses (in daltons) are asfollows: 4,243 for Rag 3,130 for Rc, and 2,584 for R. (26).Western blots (immunoblots) were performed as described

by Towbin et al. (33). BPE3, a monoclonal antibody againstpertactin (69-kDa outer membrane protein), was kindlyprovided by Roberta Shahin of the Center for BiologicsEvaluation and Research, Food and Drug Administration.

RESULTS

Growth with various iron sources. A 40% average reduc-tion in the final extent of growth was observed when B.pertussis DBP2 was grown in CSSM without the addition ofFe. Doubling times of the organism increased from 5 h in

0

0

S

00

Y.0'aa

CL

0

250

200

150

100

50

0

I t - -A

0 10 20 30 40 50

Time (hours)FIG. 1. Iron-dependent growth of DBP2 in CSSM. Symbols: A,

no Fe added; O, 10 ,uM FeSO4; 0, 10 ,uM ferric dicitrate; *, 10 ,uMhemin.

Fe-replete medium to 6 h in Fe-depleted CSSM. B. pertussiswas capable of utilizing FeSO4, ferric dicitrate, and heme asthe sole source of Fe (Fig. 1). DBP2 also grew well in thepresence of ferric pyrophosphate (data not shown). Similarresults were observed when the organism was grown inCDM (data not shown). Increasing the concentration ofadded Fe to 300 puM Fe did not enhance the rate or the finalextent of growth (data not shown).

In preliminary experiments, we observed that LF and TFprecipitated in sterile CSSM after overnight incubation at37°C. The reasons for this are unknown. The LF and TFgrowth curves were done with CDM, which has been usedextensively to study the growth of pathogenic Neisseria spp.with LF and TF (4, 16, 35, 37). B. pertussis grew with thesame doubling times in CSSM and CDM, but to a slightlyhigher final extent in CDM (data not shown). In theseexperiments, we added 10 mM NaHCO3 to the medium toensure that the Fe3+ atom remained tightly bound to LF andTF (10). We used LF or TF labeled with 1251I in theseexperiments so that leakage of the protein from the dialysistubing could be monitored. Since LF may undergo oxidativedamage as a result of radioiodination (25), we examinediodinated LF by SDS-polyacrylamide gel electrophoresis(PAGE) and autoradiography. LF labeled with lodo-beadswas not damaged by this method (data not shown). Figure 2shows the results of a typical growth curve with 1251I-labeledLF. DBP2 grew with a doubling time of approximately 7.5 hwhen the 125I-labeled LF was free in the medium. B.pertussis was capable of using LF and TF as the Fe sources,even when the protein was sequestered within dialysistubing (Fig. 2 and data not shown). A longer lag period wascommonly observed when cell-to-protein contact wasblocked, but the doubling time of the culture and the finalextent of growth remained the same. These data suggestedthat growth may be enhanced by cell-to-protein contact, butthis contact was not essential for the utilization of Fe3+ fromLF (Fig. 2) and TF (data not shown). To ensure that thedialysis bag had not leaked during the experiment, samplesfrom the '25I-labeled LF cultures were counted in a Beck-man 5500 gamma counter (Beckman Instruments, Irvine,Calif.). In each of three experiments, no appreciable radio-activity was detected in the medium when the LF wassequestered inside dialysis tubing. A control culture grownwith CDM inside dialysis tubing grew at the same rate as did

VOL. 60, 1992

on August 29, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

120 AGIATO AND DYER

0-

0

S00

40.0%-

aQ0

250

200

150

100

50

0

0.40

a0c

.00a.0

0.30

0.20 h

0.10 h

0 10 20 30 40 50 60 70

Time (hours)

FIG. 2. Growth of DBP2 with LF as the sole Fe source in CDM.Symbols: A, dialysis membrane control (no added Fe or LF); E, 10,uM FeSO4; 0, LF free in the medium; 0, LF sequestered insidedialysis membrane.

cells grown in CDM without the addition of Fe, indicatingthat exogenous Fe was not introduced on the dialysis tubingand that the dialysis tubing was not inhibiting growth in anyway (Fig. 2). This suggested that a siderophore-mediated Feacquisition system may play a major role in Fe uptake fromhost Fe-binding proteins.

Siderophore assays. The CAS assay of Schwyn andNeilands (28) detected 12 ,uM (Desferal equivalents) ofFe-binding material in the supernatant of B. pertussis grownunder Fe-depleted conditions. This Fe-binding activity ap-peared in the medium in response to Fe stress and not underFe-replete conditions, which is a common characteristic ofsiderophore biosynthesis (20). However, the CAS reagentwill also detect low-affinity Fe chelates (such as phosphate)that are not siderophores (28). The results of the Csaky assayindicated that B. pertussis DBP2 secreted 28 puM hydrox-amic acid in response to Fe deprivation. No phenolatecompound was detected using the Arnow assay (2). TheCsaky assay (7) was also used to examine the supernatantsof strains BB114 and 18323 grown in CSSM. In one typicalexperiment, BB114 excreted 80 ,uM hydroxamate and 18323excreted 70 ,uM hydroxamate.The ability of the hydroxamate excreted by strain DBP2 to

bind Fe3+ was tested by examining the difference absorptionspectra of culture supernatants after the addition of Fe3".Culture supernatant from the culture with Fe was used as thereference in this experiment. Fe-binding compounds typi-cally have an absorption maximum in the range of 400 to 500nm (19). At pH 7.6 (the pH to which SSM is buffered [30]),the difference absorption spectrum of Fe-starved culturesupernatants [containing 100 ,uM Fe(NO3)3] had a broadpeak between 400 and 500 nm, with the maximum absorption(Amax) at 450 nm (Fig. 3). Visually, culture supernatantsfrom Fe-starved DBP2 turned a faint red color when ferricnitrate was added. These observations are characteristic ofmicrobial siderophores (19). We also examined the responseof this absorbance spectrum to changes in pH. The pH offilter-sterilized culture supernatant from cells grown at pH7.6 (final pH of the culture was 7.9) was lowered in incre-ments by the addition of 0.5 M HCI. At several pH values, adifference absorption spectrum was determined. Figure 4illustrates the effect of pH on the Amax of the ferri-hydrox-amate complex in crude supernatant. The absorption peak of

0.00 L380 420 460 500 540 580 620

wavelength (nm)FIG. 3. Difference absorbance scan of spent culture supernatant

at pH 7.9 from Fe-stressed bacteria after the addition of 100 ,uMFe(NO3)3. Spent culture supernatant of Fe-replete bacteria was usedas a reference.

450 nm remained constant between pH values of 7.9 and 5.0.This peak shifted to 460 nm as the pH was lowered to 4.5 and4.0. At pH 3.5, the Amax shifted to 490 nm, and at pH 3;0 andbelow, the absorption peak had shifted to 500 nm. This shiftin absorption maximum as the pH of the medium is loweredis a property of several hydroxamate siderophores (3, 6, 29).We have-named this hydroxamate siderophore bordetellin.TF and LF receptor assay. Since we had observed sidero-

phore production, we were interested in determiningwhether B. pertussis bound TF or LF, as suggested byRedhead et al. (24). Dot blot assays were performed to testfor LF and TF receptor activity. By using this assay, bindingof the LF-AP conjugate to strain DBP2 was detected (Fig. 5).Similar results were obtained with TF-AP (data not shown).This suggested that the organism may possess a LF or TFreceptor, as suggested by Redhead et al. (24). Although thisdot blot assay is a useful crude measure of receptor activity,the dot blot assay does not faithfully reproduce the kineticsof binding of TF to the N. meningitidis TF receptor (35).Membrane alterations in response to iron availability. Sev-

eral alterations in the membrane protein profile were consis-tently noted in response to Fe availability. Fe-repressibleproteins (FeRPs) synthesized in response to Fe starvation

510

500

490

_ 480E 470

480

450

4400 1 2 3 4 5 6 7 8

pHFIG. 4. Effect of pH on the Amax of bordetellin in spent culture

supernatants.

INFECT. IMMUN.

on August 29, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

SIDEROPHORE PRODUCTION OF B. PERTUSSIS 121

FIG. 5. Dot blot assay for LF receptor. For dots 1 and 3, cellswere grown in CDM and 10 ,uM FeSO4; for dots 2 and 4, cells weregrown in CDM with no added Fe. Dots 1 and 2 were incubated withLF-AP, while dots 3 and 4 were incubated with AP substrate.DNM2 is a N. meningitidis strain.

were seen at 93, 77, and 63 kDa (Fig. 6). Fe-inducibleproteins (FeIPs), seen only in membranes from cells grownunder high Fe concentrations, were observed at 103, 72, 24,21, and 18 kDa (Fig. 6). Occasionally, we observed changesin other membrane proteins in response to Fe availability,but these changes were not reproducible. Since the 72-kDaFeIP was near the molecular size of pertactin, the 69-kDaouter membrane adhesin of B. pertussis (5), we performed aWestern blot using monoclonal antibody BPE3 (5) to deter-mine whether this FeIP was pertactin. No difference in thelevels of pertactin with respect to Fe availability was de-tected by using this assay (data not shown). The ferricdicitrate Fe transport system in E. coli is inducible by citrate(8). For this reason, we examined the membrane proteinprofiles ofB. pertussis for citrate-inducible proteins. We alsochecked for heme-inducible proteins, because cells providedwith heme as the sole Fe source seemed to have a longer lag

A +Fe -Fe

116

97

66

45

B +Fe

-97

-66

-45

-31

-21

FIG. 6. SDS-PAGE analysis of Triton X-100-insoluble mem-brane proteins from DBP2 cells grown in CSSM with or without 10,uM FeSO4. Gels containing SDS and 7.5% (A) and 15% (B)polyacrylamide were used. Solid arrows indicateFeIPs, while openarrows indicate FeRPs. Mw standards are indicated on the right sideof each gel.

1 2 3 4 5 6 7

FIG. 7. Urea-SDS-PAGE of DBP2 membranes. Lanes 1 and 3are membranes from cells grown in CSSM plus 10 ,uM FeSO4; lanes2 and 4 are membranes from cells grown in CSSM with no added Fe.Lanes 1 and 2 contain intact membranes, while lanes 3 and 4 containmembranes digested with proteinase K. Lanes 5 to 7 contain LPSfrom S. minnesota mutants Ra (lane 5), Rr (lane 6), and Re (lane 7).

phase than cells provided with other Fe sources, suggestingthat the heme uptake system may be inducible. Neithercitrate- nor heme-inducible proteins were observed (data notshown).We observed changes in the LPS in response to Fe

starvation. B. pertussis synthesizes two types of LPS (21).LPS type II has a higher phosphate content and migratesfaster in SDS-PAGE than LPS type I. The observed level oftype II LPS was severely reduced in organisms that werestressed for Fe (Fig. 7). This alteration was not due toassociation of LPS with membrane proteins, since migrationwas identical regardless of proteinase K digestion. Thissuggests that LPS phosphorylation may be inhibited by lowFe availability.

DISCUSSIONB. pertussis is capable of growth with a variety of Fe

sources. Ferrous sulfate is the Fe source incorporated intothe original formulation of SSM (30). Ferric dicitrate, ferricpyrophosphate, and heme were capable of supporting thegrowth of B. pertussis. The organism secretes a hydroxam-ate siderophore, which we named bordetellin, in response toFe deprivation. Bordetellin is probably responsible for re-moving Fe3+ from LF and TF.The ability of B. pertussis to use LF and TF as Fe sources

was not dependent upon cell-to-protein contact, as demon-strated by dialysis bag experiments in which LF or TF wasphysically separated from the organisms. A longer lag phasewas observed when direct contact between the cell and LFwas prevented, although the bacteria ultimately reached thesame final cell density. This suggests that LF binding to thecell may facilitate the removal of Fe3+ from the protein bybordetellin. Such a mechanism would presumably enhancethe efficiency of a siderophore. Recently, Menozzi et al. (17)described the purification of a 28-kDa LF-binding proteinthat may participate in enhancing the removal of Fe fromLF. Redhead et al. (24) suggested that B. pertussis possesseda single receptor which bound LF and TF; this may be theLF-binding protein identified by Menozzi et al. (17). How-ever, equilibrium binding assays to demonstrate saturable,concentration-dependent binding of 1251I-labeled LF and TFare necessary to demonstrate the presence of a TF or LFreceptor.We have reported the presence of several Fe-regulated

proteins. It is probable that some of these FeRPs play a rolein the Fe uptake system. The receptor for the ferri-borde-tellin complex is most likely Fe regulated, as are other

1 2 3 4

DBP2

DNM2

VOL. 60, 1992

-.f:.

Aii-mK----LL.

on August 29, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

122 AGIATO AND DYER

siderophore receptors (20). We observed several proteinswhose synthesis is enhanced by Fe-replete conditions. Al-though the function of these proteins is unknown, onepossible role may be in Fe storage. The Fe-dependentalterations in the LPS of the organism may also play somerole in the pathogenesis of pertussis.From the results of this study, we can propose a model for

the functioning of the Fe uptake system of B. pertussisduring the process of disease in the human host. In the earlystages of infection, B. pertussis grows on the upper respira-tory mucosa. We infer that sufficient heme may be presenton this mucosal surface to support the growth of B. pertus-sis, since H. influenzae is often found at this site. H.influenzae requires heme (or at least porphyrin) for growth(11). However, the concentration of heme present in thesemucosal secretions is probably low, and its origin is un-known. Porphyrin may be released by other organismsgrowing in the mixed microbial environment of the upperrespiratory tract. LF is also present in these respiratorymucosal secretions and maintains free Fe at low levels.Therefore, it is likely that B. pertussis uses both heme andLF as Fe sources during the early part of infection. Weobserved that organisms grown with heme as the sole Fesource had the same doubling time as cells grown withFeSO4 and Fe dicitrate. When LF was the only Fe source,however, the doubling times increased by 2.5 h. This sug-gests that heme may be more readily used as an Fe source onthe respiratory mucosum, and the contribution of LF-Fe tothe growth of the organism may be somewhat less. As theinfection progresses, B. pertussis descends into the lowerrespiratory tract. This is accompanied by significant inflam-mation in the lower airways. In patients with chronic bron-chitis, the LF levels in bronchiolavage fluids were 10-foldhigher than normal (32). TF levels increased fourfold in theselavage fluids as a result of transudation of plasma compo-nents (32). We suggest that during the latter stages ofinfection B. pertussis is growing in an Fe-rich environment inwhich LF and TF are the primary Fe sources. Paroxysmal-stage inflammation will also lower the pH of the respiratorymucosal secretions. This will tend to increase even furtherthe availability of Fe from LF and TF, because the Fe onthese proteins becomes labile at lower pH (1). These factorsmay increase Fe availability dramatically during the latterstages of the disease. We observed that certain membraneproteins are produced by B. pertussis under high Fe avail-ability. It is possible that these FeIPs may represent proteinsthat are important for the pathogenesis of the latter stages ofinfection by B. pertussis.

ACKNOWLEDGMENTS

This work was supported in part by grant S-90-1 from theDiamond Research Fund of the Graduate Student Association of theState University ofNew York at Buffalo (L.-A.A.) and by BiologicalResearch Support grant 150-937-OK from the State University ofNew York at Buffalo (D.W.D.).We would like to thank Murray Ettinger and Kathy Bethin for

performing the atomic absorption measurements, Mark O'Brien forhelpful comments during the preparation of the manuscript, andRickie Duffy for artwork and photography.

REFERENCES1. Aisen, P., and A. Leibman. 1972. Lactoferrin and transferrin: a

comparative study. Biochim. Biophys. Acta 257:314-323.2. Arnow, E. 1937. Colorimetric determination of the components

of 3,4-dihydroxylphenylalanine-tyrosine mixtures. J. Biol.Chem. 118:531-537.

3. Atkin, C. L., and J. B. Neilands. 1968. Rhodotorulic acid, a

diketopiperazine dihydroxamic acid with growth factor-likeactivity. I. Isolation and characterization. Biochemistry 7:3734-3739.

4. Blanton, K. J., G. D. Biswas, J. Tsai, J. Adams, D. W. Dyer, M.Davis, G. G. Koch, P. K. Sen, and P. F. Sparling. 1990. Geneticevidence that Neisseria gonorrhoeae produces specific recep-tors for transferrin and lactoferrin. J. Bacteriol. 172:5225-5235.

5. Brennan, M. J., Z. M. Li, J. L. Cowell, M. E. Bishner, A. C.Steven, P. Novotny, and C. R. Manclark. 1988. Identification ofa 69-kilodalton nonfimbrial protein as an agglutinogen of Borde-tella pertussis. Infect. Immun. 56:3189-3195.

6. Byers, B. R., M. V. Powell, and C. E. Lankford. 1967. Iron-chelating hydroxamic acid (schizokinen) active in initiation ofcell division in Bacillus megaterium. J. Bacteriol. 93:286-294.

7. Csaky, T. Z. 1948. On the estimation of bound hydroxylamine inbiological materials. Acta Chem. Scand. 2:450-454.

8. Frost, G. E., and H. Rosenberg. 1973. The inducible citratedependent iron-transport system in Escherichia coli K-12. Bio-chim. Biophys. Acta 330:90-101.

9. Gorringe, A. R., G. Woods, and A. Robinson. 1990. Growth andsiderophore production by Bordetella pertussis under iron re-stricted conditions. FEMS Microbiol. Lett. 66:101-106.

10. Graham, G. A., and G. W. Bates. 1977. Factors influencing therate of release of iron from Fe3+-transferrin-CO327 p. 273-290.In E. B. Brown, P. A. J. Fielding, and R. R. Crichton (ed.),Proteins of iron metabolism. Grune and Stratton, New York.

11. Herrington, D. A., and P. F. Sparling. 1985. Haemophilusinfluenzae can use human transferrin as a sole source forrequired iron. Infect. Immun. 48:248-251.

12. Hitchcock, P. J., and T. M. Brown. 1983. Morphological heter-ogeneity among Salmonella lipopolysaccharide chemotypes insilver-stained polyacrylamide gels. J. Bacteriol. 154:269-277.

13. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

14. Mandrell, R. E., H. Schneider, W. D. Zollinger, M. A. Apicella,and J. M. Grifiss. 1985. Characterization of mouse monoclonalantibodies specific for gonococcal lipopolysaccharides, p. 379-384. In 0. F. Schoolnik, G. F. Brooks, S. Falkow, C. E. Frasch,J. S. Knapp, J. A. McCutchan, and S. A. Morse (ed.), Thepathogenic neisseriae. Proceedings of the Fourth InternationalSymposium. American Society for Microbiology, Washington,D.C.

15. Markwell, M., S. M. Haas, L. L. Bieber, and N. E. Tolbert. 1978.A modification of the Lowry procedure to simplify proteindetermination in membrane and lipoprotein samples. Anal.Biochem. 87:206-210.

16. McKenna, W. R., P. A. Mickelsen, P. F. Sparling, and D. W.Dyer. 1988. Iron uptake from lactoferrin and transferrin byNeisseria gonorrhoeae. Infect. Immun. 56:785-791.

17. Menozzi, F. D., C. Gantiez, and C. Locht. 1991. Identificationand purification of transferrin- and lactoferrin-binding proteinsof Bordetella pertussis and Bordetella bronchiseptica. Infect.Immun. 59:3982-3988.

18. Morton, D. J., and P. Williams. 1990. Siderophore independentacquisition of transferrin-bound iron by Haemophilus influenzaeinfluenzae type b. J. Gen. Microbiol. 136:927-933.

19. Neilands, J. B. 1981. Microbial iron compounds. Annu. Rev.Biochem. 50:715-731.

20. Neilands, J. B. 1982. Microbial envelope proteins related to iron.Annu. Rev. Microbiol. 36:285-309.

21. Peppler, M. 1984. Two physically and serologically distinctlipopolysaccharide profiles in strains of Bordetella pertussis andtheir phenotypic variants. Infect. Immun. 43:224-232.

22. Redhead, K. 1983. Variability in the surface exposure of theouter membrane of Bordetella pertussis. J. Gen. Microbiol.17:35-39.

23. Redhead, K., and T. Hill. 1991. Acquisition of iron fromtransferrin by Bordetella pertussis. FEMS Microbiol. Lett.77:303-308.

24. Redhead, K., T. Hill, and H. Chart. 1987. Interaction of trans-ferrin and lactoferrin with outer membrane of B. pertussis. J.Gen. Microbiol. 133:891-898.

INFECT. IMMUN.

on August 29, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

SIDEROPHORE PRODUCTION OF B. PERTUSSIS 123

25. Rosenmund, A., C. Kuyas, and A. Haeberli. 1986. Oxidativeradioiodination damage to human lactoferrin. Biochem. J. 240:239-245.

26. Schneider, H., T. L. Hale, W. D. Zoilinger, R. C. Seid, C. A.Hammack, and M. Griffiss. 1984. Heterogeneity of molecularsize and antigenic expression within lipopolysaccharides ofindividual strains of Neisseria gonorrhoeae and Neisseria men-

ingitidis. Infect. Immun. 45:544-549.27. Schryvers, A. B., and L. J. Morris. 1988. Identification and

characterization of the transferrin receptor from N. meningiti-dis. Mol. Microbiol. 2:281-288.

28. Schwyn, B., and J. B. Neilands. 1987. Universal chemicaldetection and determination of siderophores. Anal. Biochem.160:47-56.

29. Seifter, S., P. M. Gallop, S. Michaels, and E. Meilman. 1960.Analysis of hydroxamic acids and hydrazines: preparation andproperties of dinitrophenyl derivatives of hydroxamic acids,oxamines, hydrazines, and hydrazones. J. Biol. Chem. 235:2613-2618.

30. Stainer, D. W., and M. J. Scholte. 1971. A simple chemicallydefined medium for the production of phase I Bordetella pertus-sis. J. Gen. Microbiol. 63:211-220.

31. Stookey, L. L. 1970. Ferrozine-a new spectrophotometricassay for iron. Anal. Chem. 42:779-781.

32. Thompson, A. B., T. Bohling, F. Payvandi, and S. I. Rennard.1990. Lower respiratory tract lactoferrin and lysozyme ariseprimarily in the airways and are associated with chronic bron-chitis. J. Lab. Clin. Med. 115:148-158.

33. Towbin, H., T. Staehelin, and J. Gordon. 1979. Electrophoretictransfer of proteins from polyacrylamide gels to nitrocellulosesheets: procedure and some applications. Proc. Natl. Acad. Sci.USA 76:4350-4354.

34. Tsai, C. M., and C. Frasch. 1981. A sensitive silver stain fordetecting lipopolysaccharides in polyacrylamide gels. Anal.Biochem. 119:115-119.

35. Tsai, J., D. W. Dyer, and P. F. Sparling. 1988. Loss oftransferrin receptor activity in Neisseria meningitidis correlateswith the inability to use transferrin as an iron source. Infect.Immun. 56:3132-3138.

36. Weinberg, E. D. 1978. Iron and infection. Microbiol. Rev.42:45-66.

37. West, S. E. H., and P. F. Sparling. 1987. Aerobactin utilizationby Neisseria gonorrhoeae and the cloning of a genomic DNAfragment that complements Escherichia colifhuB mutations. J.Bacteriol. 169:3414-3421.

38. Wray, W., T. Boulikas, V. P. Wray, and R. Hancock. 1981.Silver staining of proteins in polyacrylamide gels. Anal. Bio-chem. 118:197-203.

VOL. 60, 1992

on August 29, 2020 by guest

http://iai.asm.org/

Dow

nloaded from